struct radeon_winsys_cs *cs = sctx->b.gfx.cs;
unsigned sh_base_reg = sctx->shader_userdata.sh_base[PIPE_SHADER_VERTEX];
bool render_cond_bit = sctx->b.render_cond && !sctx->b.render_cond_force_off;
+ uint32_t index_max_size = 0;
+ uint64_t index_va = 0;
if (info->count_from_stream_output) {
struct r600_so_target *t =
assert(!"unreachable");
return;
}
+
+ index_max_size = (ib->buffer->width0 - ib->offset) /
+ ib->index_size;
+ index_va = r600_resource(ib->buffer)->gpu_address + ib->offset;
+
+ assert(index_va % 2 == 0);
+
+ radeon_add_to_buffer_list(&sctx->b, &sctx->b.gfx,
+ (struct r600_resource *)ib->buffer,
+ RADEON_USAGE_READ, RADEON_PRIO_INDEX_BUFFER);
}
if (!info->indirect) {
sctx->last_sh_base_reg = sh_base_reg;
}
} else {
+ uint64_t indirect_va = r600_resource(info->indirect)->gpu_address;
+
+ assert(indirect_va % 8 == 0);
+
si_invalidate_draw_sh_constants(sctx);
+ radeon_emit(cs, PKT3(PKT3_SET_BASE, 2, 0));
+ radeon_emit(cs, 1);
+ radeon_emit(cs, indirect_va);
+ radeon_emit(cs, indirect_va >> 32);
+
radeon_add_to_buffer_list(&sctx->b, &sctx->b.gfx,
(struct r600_resource *)info->indirect,
RADEON_USAGE_READ, RADEON_PRIO_DRAW_INDIRECT);
}
- if (info->indexed) {
- uint32_t index_max_size = (ib->buffer->width0 - ib->offset) /
- ib->index_size;
- uint64_t index_va = r600_resource(ib->buffer)->gpu_address + ib->offset;
-
- radeon_add_to_buffer_list(&sctx->b, &sctx->b.gfx,
- (struct r600_resource *)ib->buffer,
- RADEON_USAGE_READ, RADEON_PRIO_INDEX_BUFFER);
-
- if (info->indirect) {
- uint64_t indirect_va = r600_resource(info->indirect)->gpu_address;
-
- assert(indirect_va % 8 == 0);
- assert(index_va % 2 == 0);
- assert(info->indirect_offset % 4 == 0);
+ if (info->indirect) {
+ unsigned di_src_sel = info->indexed ? V_0287F0_DI_SRC_SEL_DMA
+ : V_0287F0_DI_SRC_SEL_AUTO_INDEX;
- radeon_emit(cs, PKT3(PKT3_SET_BASE, 2, 0));
- radeon_emit(cs, 1);
- radeon_emit(cs, indirect_va);
- radeon_emit(cs, indirect_va >> 32);
+ assert(info->indirect_offset % 4 == 0);
+ if (info->indexed) {
radeon_emit(cs, PKT3(PKT3_INDEX_BASE, 1, 0));
radeon_emit(cs, index_va);
radeon_emit(cs, index_va >> 32);
radeon_emit(cs, PKT3(PKT3_INDEX_BUFFER_SIZE, 0, 0));
radeon_emit(cs, index_max_size);
+ }
- if (sctx->b.family < CHIP_POLARIS10) {
- radeon_emit(cs, PKT3(PKT3_DRAW_INDEX_INDIRECT, 3, render_cond_bit));
- radeon_emit(cs, info->indirect_offset);
- radeon_emit(cs, (sh_base_reg + SI_SGPR_BASE_VERTEX * 4 - SI_SH_REG_OFFSET) >> 2);
- radeon_emit(cs, (sh_base_reg + SI_SGPR_START_INSTANCE * 4 - SI_SH_REG_OFFSET) >> 2);
- radeon_emit(cs, V_0287F0_DI_SRC_SEL_DMA);
- } else {
- radeon_emit(cs, PKT3(PKT3_DRAW_INDEX_INDIRECT_MULTI, 8, render_cond_bit));
- radeon_emit(cs, info->indirect_offset);
- radeon_emit(cs, (sh_base_reg + SI_SGPR_BASE_VERTEX * 4 - SI_SH_REG_OFFSET) >> 2);
- radeon_emit(cs, (sh_base_reg + SI_SGPR_START_INSTANCE * 4 - SI_SH_REG_OFFSET) >> 2);
- radeon_emit(cs, 0); /* draw_index */
- radeon_emit(cs, 1); /* count */
- radeon_emit(cs, 0); /* count_addr -- disabled */
- radeon_emit(cs, 0);
- radeon_emit(cs, 16); /* stride */
- radeon_emit(cs, V_0287F0_DI_SRC_SEL_DMA);
- }
+ if (!sctx->screen->has_draw_indirect_multi) {
+ radeon_emit(cs, PKT3(info->indexed ? PKT3_DRAW_INDEX_INDIRECT
+ : PKT3_DRAW_INDIRECT,
+ 3, render_cond_bit));
+ radeon_emit(cs, info->indirect_offset);
+ radeon_emit(cs, (sh_base_reg + SI_SGPR_BASE_VERTEX * 4 - SI_SH_REG_OFFSET) >> 2);
+ radeon_emit(cs, (sh_base_reg + SI_SGPR_START_INSTANCE * 4 - SI_SH_REG_OFFSET) >> 2);
+ radeon_emit(cs, di_src_sel);
} else {
+ radeon_emit(cs, PKT3(info->indexed ? PKT3_DRAW_INDEX_INDIRECT_MULTI :
+ PKT3_DRAW_INDIRECT_MULTI,
+ 8, render_cond_bit));
+ radeon_emit(cs, info->indirect_offset);
+ radeon_emit(cs, (sh_base_reg + SI_SGPR_BASE_VERTEX * 4 - SI_SH_REG_OFFSET) >> 2);
+ radeon_emit(cs, (sh_base_reg + SI_SGPR_START_INSTANCE * 4 - SI_SH_REG_OFFSET) >> 2);
+ radeon_emit(cs, 0); /* draw_index */
+ radeon_emit(cs, 1); /* count */
+ radeon_emit(cs, 0); /* count_addr -- disabled */
+ radeon_emit(cs, 0);
+ radeon_emit(cs, 16); /* stride */
+ radeon_emit(cs, di_src_sel);
+ }
+ } else {
+ if (info->indexed) {
index_va += info->start * ib->index_size;
radeon_emit(cs, PKT3(PKT3_DRAW_INDEX_2, 4, render_cond_bit));
radeon_emit(cs, (index_va >> 32UL) & 0xFF);
radeon_emit(cs, info->count);
radeon_emit(cs, V_0287F0_DI_SRC_SEL_DMA);
- }
- } else {
- if (info->indirect) {
- uint64_t indirect_va = r600_resource(info->indirect)->gpu_address;
-
- assert(indirect_va % 8 == 0);
- assert(info->indirect_offset % 4 == 0);
-
- radeon_emit(cs, PKT3(PKT3_SET_BASE, 2, 0));
- radeon_emit(cs, 1);
- radeon_emit(cs, indirect_va);
- radeon_emit(cs, indirect_va >> 32);
-
- if (sctx->b.family < CHIP_POLARIS10) {
- radeon_emit(cs, PKT3(PKT3_DRAW_INDIRECT, 3, render_cond_bit));
- radeon_emit(cs, info->indirect_offset);
- radeon_emit(cs, (sh_base_reg + SI_SGPR_BASE_VERTEX * 4 - SI_SH_REG_OFFSET) >> 2);
- radeon_emit(cs, (sh_base_reg + SI_SGPR_START_INSTANCE * 4 - SI_SH_REG_OFFSET) >> 2);
- radeon_emit(cs, V_0287F0_DI_SRC_SEL_AUTO_INDEX);
- } else {
- radeon_emit(cs, PKT3(PKT3_DRAW_INDIRECT_MULTI, 8, render_cond_bit));
- radeon_emit(cs, info->indirect_offset);
- radeon_emit(cs, (sh_base_reg + SI_SGPR_BASE_VERTEX * 4 - SI_SH_REG_OFFSET) >> 2);
- radeon_emit(cs, (sh_base_reg + SI_SGPR_START_INSTANCE * 4 - SI_SH_REG_OFFSET) >> 2);
- radeon_emit(cs, 0); /* draw_index */
- radeon_emit(cs, 1); /* count */
- radeon_emit(cs, 0); /* count_addr -- disabled */
- radeon_emit(cs, 0);
- radeon_emit(cs, 16); /* stride */
- radeon_emit(cs, V_0287F0_DI_SRC_SEL_AUTO_INDEX);
- }
} else {
radeon_emit(cs, PKT3(PKT3_DRAW_INDEX_AUTO, 1, render_cond_bit));
radeon_emit(cs, info->count);
radeon_emit(cs, V_0287F0_DI_SRC_SEL_AUTO_INDEX |
- S_0287F0_USE_OPAQUE(!!info->count_from_stream_output));
+ S_0287F0_USE_OPAQUE(!!info->count_from_stream_output));
}
}
}
if (sctx->b.flags)
si_mark_atom_dirty(sctx, sctx->atoms.s.cache_flush);
+ /* Add buffer sizes for memory checking in need_cs_space. */
+ if (sctx->emit_scratch_reloc && sctx->scratch_buffer)
+ r600_context_add_resource_size(ctx, &sctx->scratch_buffer->b.b);
+ if (info->indirect)
+ r600_context_add_resource_size(ctx, info->indirect);
+
si_need_cs_space(sctx);
+ /* Since we've called r600_context_add_resource_size for vertex buffers,
+ * this must be called after si_need_cs_space, because we must let
+ * need_cs_space flush before we add buffers to the buffer list.
+ */
+ if (!si_upload_vertex_buffer_descriptors(sctx))
+ return;
+
/* Emit states. */
mask = sctx->dirty_atoms;
while (mask) {